Document Type


Date of Award


Degree Name

Master of Science (MS)


Systems Engineering

First Advisor

Sayama, Hiroki

Second Advisor

Laramee, Craig

Third Advisor

Lewis, Harold


Stigmergy has long been studied and recognized as an effective system for self-organization among social insects. Through the use of chemical agents known as pheromones, insect colonies are capable of complex collective behavior often beyond the scope of an individual agent. In an effort to develop human-made systems with the same robustness, scientists have created artificial analogues of pheromone-based stigmergy, but these systems often suffer from scalability and complexity issues due to the problems associated with mimicking the physics of pheromone diffusion. In this thesis, an alternative stigmergic framework called 'Genetic Stigmergy' is introduced. Using this framework, agents can indirectly share entire behavioral algorithms instead of pheromone traces that are limited in information content. The genetic constructs used in this framework allow for new avenues of research, including real-time evolution and adaptation of agents to complex environments. As a nascent test of its potential, experiments are performed using genetic stigmergy as an indirect communication framework for a simulated swarm of robots tasked with mapping an unknown environment. The robots are able to share their behavioral genes through environmentally distributed Radio-Frequency Identification cards. It was found that robots using a schema encouraging them to adopt lesser used behavioral genes (corresponding with novelty in exploration strategies) can generally cover more of an environment than agents who randomly switch their genes, but only if the environmental complexity is not too high. While the performance improvement is not statistically significant enough to clearly establish genetic stigmergy as a superior alternative to pheromonal-based artificial stigmergy, it is enough to warrant further research to develop its potential.